Dilute single-crystal electron paramagnetic resonance study of bis

Jeffrey L. Petersen, and Lawrence F. Dahl. J. Am. Chem. Soc. , 1975 ... Alexander F. R. Kilpatrick , Jennifer C. Green , and F. Geoffrey N. Cloke. Org...
0 downloads 0 Views 875KB Size
6416

Dilute Single-Crystal Electron Paramagnetic Resonance Study of Bis( cyclopentadieny1)vanadium Pentasulfide, V(q5-C5H5)2S5. The Coup de Grace to the Ballhausen-Dah1 Bonding Model Applied to M( q5-C5H5)2L2-TypeComplexes Jeffrey L. Petersen and Lawrence F. DahI* Contributionfrom the Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706. Received February 24, 1975

Abstract: A dilute single-crystal EPR investigation of V(qS-C=,H5)2S5 doped in the crystal lattice of the diamagnetic Ti($CsH5)2Ss host has provided the first quantitative determination of the relarive metal orbital character and the directional properties of the unpaired electron in a V(1V) V(q5-C~H5)2L2complex. Based upon the principal orthogonal coordinate system of the hyperfine coupling tensor being oriented within experimental error with respect to the pseudo-C2, geometry of the V(q5-C5H5)2S2 fragment such that the x axis lies along the S-V-S bisector with the y axis perpendicular to the VS2 plane and the z axis perpendicular to the plane bisecting the VS2 part, a detailed analysis of the anisotropy of the hyperfine interaction of the unpaired electron with the 51Vnucleus shows clearly that the unpaired electron resides primarily on the vanadium atom in an al-type M O mainly composed of 3d,z with a small but significant amount of 3dX+z and virtually no 4s character. These EPR results thereby provide convincing evidence that the widely utilized Ballhausen-Dah1 (qualitative) model is not valid for d’ and d2 M(IV) M(q5-C5H5)2L2 compounds and additionally indicate that the subsequent alternative Alcock (qualitative) model (which arbitrarily assumes the d1 and d2 electrons to occupy solely a dr2 AO) is not an adequate representation of this MO. The directional properties of the metal components of the M O are compatible with the observed decrease in the L-M-L bond angle upon its occupation. The average principal values of the g tensor and hyperfine tensor for the two nonequivalent magnetic sites are gx = 1.9964, g, = 1.9997, and g, = 1.9689 and T, = (-)66.6 G, T, = (-)111.3 G , and T , = (-)23.5 G; these two tensors were determined to be noncoincident (presumably due at least in part to significant deviations of the vanadium molecule of crystallographic site symmetry C1 from C2a symmetry). The small spin-orbit coupling X (estimated to be only 30 cm-I) is in accord with the principal components of g not differing appreciably from the free-electron value. The calculated value of 93.2 G for P expectedly compares favorably with that between a Vo and V + system.

Two qualitative representations have been proposed to represent the bonding in M(v5-C5H5)2L2-type complexes. Based solely on N M R results2 Ballhausen and Dah13 in 196 1 formulated a (back-of-the-envelope) representation involving the use of hybrid metal orbitals to describe the bonding in protonated sandwich compounds. For Mo($C5H5)2H2 this model suggested that the d2 Mo(1V) electrons are located in a sterically active orbital directed between the two metal-coordinated hydrogens. This B-D representation has been widely accepted4 and generalized to other d’ and d 2 M(1V) M(v5-C5H5)2L2-type complexes containing nonhydridic L ligands. An alternative description which retains the basic hybrid orbital features of the B-D approach was proposed in 1967 by Alcock5 on the basis of his concluding from an X-ray diffraction study of Re(q5-C5H5)(v4-C5H5CH3)(CH3)2 that the acute H3CRe-CH3 bond angle of 75.8 (1.3)’ indicated the nonvalidity of the B-D model in its placement of an occupied orbital between the methyl ligands. Instead, Alcock5 suggested with the rather bulky methyl groups that it is more satisfactory to have the lone pair in a metal orbital primarily directed normal to the plane bisecting the H3C-Re-CH3 bond angle. Alcocks also proposed that Mo(v5-C5H5)2H2 might have an analogous electronic structure, although he pointed out that from repulsion considerations the smaller hydrogen atoms might make the B-D structure more stable. The major difference in these two bonding models lies in the directional character of this presumed sterically active “ninth” orbital which for Mo(v5-C5H5)2H2 contains the d2 electrons. The outcome of the X-ray diffraction studies on several M(v5-C5H5)2L2-type complexes (Le., M(v5-C5H5)2(SC6H5)2 and M(v5-C5H5)2S5 where M = Ti, V), carried Journal of the American Chemical Society

/

97:22

/

out as operational tests of the B-D description, has been previously reported.6-8 These results together with those obtained by Green, Prout, and coworkers9 for second-row transition metals reveal that the L-M-L bond angle decreases in going from a do (Ti, Zr) to a d’ (V, Nb) to a d2 (Mo) system. This opposite trend in the behavior of the L-M-L bond angles with that necessitated from electron: pair repulsion arguments via the B-D model was taken as strong evidence for its invalidity when applied to M($C5H5)2L2-type complexes. In addition, Green et al.9a concluded that their crystallographic data supported the Alcock model for M(v5-C5H5)2Lz molecules; they also formulated a qualitative MO bonding description (incorporating features of both the B-D and Alcock models) for M($C5H5)2Hn ( n = 1, 2, 3) and (q5-C5H5)2M-(~2-L)2-M’Ld systems. Although considerable structural data have been obtained for M(q5-CsH5)2L2-type complexes, the metal orbital character and spatial arrangement of the “ninth” orbital in these complexes remained without any quantitative foundation. Both of the Ballhausen-Dah1 and Alcock bonding schemes are based solely on the hybridization of the metal orbitals along specific directions to maximize overlap with the ligands. The main objective of the work presented here was t o ascertain the specific nature of the “ninth” metal orbital from dilute single-crystal EPR measurements conducted on an appropriate paramagentic d’ V(1V) V(v5-C5H5)2Lz molecule. In this particular paper,lO.ll we describe the results and interpretation of an EPR investigation of V(v5CsH5)2Ss (containing the bidentate Ss2- ligand) doped in the crystal lattice of the diamagnetic host Ti(v5-CsHs)2Ss. A detailed analysis of the principal values and directions of the hyperfine coupling tensor (T), made possible from the

October 29, 1975

6417

2900

3200

3400

3300

MAGNETIC

FIELD

3500

3600

3 7 0 0 3900

(GAUSS1

Figure 1. EPR spectra of V($-CsH5)2S5 diluted in the crystal lattice of T ~ ( $ - C ~ H S )with ~ S ~(a) the a axis perpendicular to the direction of the magnetic field, and (b) the b axis perpendicular to the direction of the magnetic field H ~0

hyperfine interaction of the unpaired electron with the ” V nucleus ( I = 7h for 5 1 V ,99.8%), has been utilized to extract the metal orbital character and directional properties of the MO containing the unpaired electron.

HYPERf I N E ‘ \ : ~ ~ ~ ~ N a nNUCLEAR 6 ZEEYAN

aMI:ti

aMs:ti

aM,:o

aM,:ti

QUADRUPOLE

Figure 2. Energy-lever diagrams of spin states Irn,rnl) for S = ‘h, I = 7h.

Experimental Section In order to reduce effectively the exchange interaction between paramagnetic sites in a single-crystal EPR study, it is necessary to dilute the paramagnetic species into the crystal lattice of an isostructural diamagnetic host material. An appropriate pair of complexes for such an experiment is V(qS-C5H5)2S5 and Ti($‘CsHs)&, which have been shown by X-ray diffraction*.’*to possess nearly identical molecular structures. Single crystals of ca. 0.2% V(qS-CsH5)2Ss diluted in the crystal lattice of Ti(q5-CsHs)2S5 were grown from a DMF solution by slow evaporation of the solvent. The EPR measurements were made on several of these deep-red crystals which were found from X-ray oscillation and Weissenberg photographs to possess monoclinic Laue symmetry C2h-2/m with 0 = 93’. From the X-ray photographs it was also determined that the lattice parameters of the doped crystals did not vary by more than 1% from those reported by Epstein, Bernal, and Kopf12 for single crystals of monoclinic Ti($-CsHs)&, which were also grown from a DMF solution. These parallelepiped crystals were mounted with a water-soluble glue on the end of 0.5-mm Lindemann capillary tubes, which in turn were each glued to a larger 5-mm quartz tube that fit into the holder of a Varian single-crystal goniometer. The crystals were oriented along the a , b, and c crystallographic axesi3 to within OS’, 0.2’, and 2.0°, respectively. The doped crystals were placed individually in a Varian cylindrical microwave cavity between the pole faces of a Varian 1541. rotating magnet such that the spindle axis of the goniometer was perpendicular to the direction of the magnetic field and the plane of rotation. The orientation of the crystal with respect to the magnetic field direction was changed by a counterclockwise rotation of the magnet. A 1000-G scan of I-hr duration was obtained for every IO0 rotation about each of the three crystallographic directions. For particular orientations where the two observed overlapping spectra (due to the two magnetically nonequivalent sites in the host lattice) were nearly equivalent, spectra were obtained at 5’ intervals. From symmetry considerations of the crystal structure of the host monoclinic Ti($-CsHs)& lattice, two magnetically nonequivalent orientations of the V(s5-CsH5)2S5 molecules are present. Two overlapping eight-line spectra were observed except when the b axis was perpendicular to or parallel with the direction of the magnetic field. For each of these particular orientations only one eight-line spectrum was observed. Figure 1 shows two representative EPR spectra obtained for the dilute single-crystals of V($-C5Hs)&; Figure la shows the presence of two overlapping eight-line spectra taken with the a axis perpendicular to the magnetic field, while Figure 1 b has only one eight-line spectrum obtained with the b axis perpendicular to the direction of the magnetic field.

Petersen, Dah1

A closer look a t Figure la reveals that between the eight more intense “allowed” resonance lines there are seven pairs of weaker resonance lines which represent “forbidden” transitions. These I4 lines are a consequence of the anisotropic part of the hyperfine coupling interaction which provides a mechanism for the mixing of nuclear spin states.14 This effect leads to transitions in the EPR spectra where both the electron and nuclear spin may change simultaneously (Le., Ams = & I , Am1 = & I ) . An examination of the spectra containing these weaker resonance lines shows that the spacing between two “forbidden” transitions comprising a “forbidden pair” increases slightly with the magnetic field strength. This variation has been used to estimate the magnitude of the nuclear quadrupole interaction.Is For nuclei such as 5 ’ V where the nuclear spin quantum number I > I,$, a nuclear quadrupole interaction described by 1.Q.Z is possible. However, since the variation in spacing between airs of “forbidden” lines ranges from ca. 1 to 3 G, the quadrupo e interaction for V ( $ - C S H ~ ) ~isS reasonably ~ small (estimated to be ca. = 0.1-0.2 G). Figure 2 represents the energylevel diagram for a S = ’,$,I = 7h system and illustrates the origin of the eight “allowed” transitions and the 14 “forbidden” transitions. Compared to the electron Zeeman and hyperfine interactions, the nuclear Zeeman and nuclear quadrupole terms are several orders of magnitude smaller for V($-C5Hs)lSs.

T

la

Data Analysis The initial g value for each spectrum was calculated from the average of the magnetic fields of the eight hyperfine lines, Ha,, from the relationship g,, = h u / P H , , . Since second-order effects make the smallest contributions to the two inner-most hyperfine lines, the experimental hyperfine splitting value for each V(q5-CsH5)2S5 spectrum was calculated from the difference between the magnetic field strengths for these two hyperfine lines. Least-square fits to the experimental data were made with a three-parameter equation y i = P cos2 Bi

+ Q sin2 8, - 2R sin 8, cos 8,

where yi = g2 or g2T2 and 8, is the angle of rotation. From the “best” values of P , Q, and R obtained for the three (nearly) orthogonal sets of crystal orientations, the matrix elements of the g2 and K 2 = g 2 T 2 matrices were determined.I6 The g 2 matrix was diagonalized to obtain the principal directions and principal components of the g tensor. hg2= c g g 2 C ,

/

EPR Study of Bis(cyclopentadieny1)uanadium Pentasulfi‘de

6418

72 70 68

y

/

::[ 62 62 60 58

alH

b lH

clH

Figure 3. Plots of experimental g values and hyperfine coupling constants, T , vs. the angle of rotation, 8 ~ for . the three nearly orthogonal sets’l of EPR data obtained for V(q5-CsH5)2Ss diluted in the crystal lattice of Ti(qS-C5Hs)&. The solid lines represent the best fit of the experimental data (open symbols) to a three-parameter equation, P cos2 0, Q sin2 8, - 2R sin 0, cos 8,.

+

The diagonal elements of Ag,which are the principal values of g, are simply the square roots of the diagonal elements of the Ag2 matrix. All of the diagonal elements of Ag were assumed to be positive. The inverse of g was then determined and used to calculate the T 2 matrix from the relationship, T2 = g - ’ K 2 g - I . The T 2 matrix was then diagonalized to determine the principal values and directions of T. AT’

=

CTT~CT

Since the diagonal elements of AT are equal to the square roots of the diagonal elements of AT^, the principal values of T are known to within a sign. The orientations of the principal axes of g and T with respect to the laboratoryfixed crystallographic axial system were extracted from the transformation matrices C, and C T , respectively, and reexpressed in terms of the Eulerian angles a, 0, and Q . I 7 A second-order correction to the initial g values was performed for each spectrum. For the spin Hamiltonian X = P S - g H S.T.1, which contains the two principal magnetic interactions observed for V(q5-C5H5)2S5, the energies through second-order have been found by the solution for the eigenvalues of X via the usual perturbation theory techniques. Weil et a1.I8 have derived the following expression for the “allowed” transition energies for a S = ‘/2, I = n / 2 system.

+

hv = E ( 1 / 2 , m l ) - E ( - 1 / 2 ,

[ ( I ( I + 1)

m ~ =)

2

Molecule 1

Molecule 2 KZ

gz

3.94562 17960.6 3.94543 3.97809 34821.4 3.97810 -0.0326216 0.0327681 -20887.7 0.00032 0.00034 93 0.62 0.78 “T 3.93655 15231.1 3.93655 pb 3.94565 17860.2 3.94565 Qb -0.0447060 -0.0447060 3231.37 Rb 0.00056 0.00056 “g 0.45 0.45 “T 3.97719 3.98135 36126.5 PC 3.93733 16180.3 3.93821 Qc -0.029703 0.033 1762 4059.55 RC 0.00019 0.00022 “g 0.32 0.30 “T 1.9963 1.9965 gx 1.9999 1.9998 gY 1.9691 1.9686 gz 1.9884 1.9883 gav -59.47 58.15 6 e 51.81 128.97 125.83 i -61.58 Tx (-)66.7 G (-)66.4 G (-)112.0 G (-)110.7 G (-)23.6G (-)23.4 G Tav (-)67.4 G (-)66.8 G @ 33.04 -31.77 e 140.79 39.45 i 28.81 207.12 Pa Qa Ra

KZ

18129.3 34813.1 20654.0 15231.0 17860.2 3231.37 33400.9 16126.1 -3743.20

2

- m12)/21 + [ ( f 3 1 2 + t 3 2 2 ~ 3 3 mi2J 1

where t = T2. If C is set equal to the contents in brackets, the resulting quadratic expression in g can be obtained. 0 = g 2 + ((t33)i/2mi- u ) ( h / P H ) g + 1 / 2 ( l ~ / b H ) ~ C

From the components of T at the appropriate crystal orientation and the values of the corresponding experimental hyperfine line positions, a set of eight g values was computed for each spectrum. The final g value was designated as the average of the corrected g values calculated for the two inner-most hyperfine lines. The previously described (leastsquares fit)-routine was then repeated. The final components of the g2 and the K 2 matrices for both molecules of V(q5-CsH5)2S5 are listed in Table I . The results of the least-squares fits of the EPR data are illustrated in Figure 3, which presents plots of gexptl vs. 0 R and Texptlvs. 0 R . Journal of the American Chemical Society

Table I. Final Analysis of SingleCrystal EPR Data Obtained for the Two Magnetically Nonequivalent V(qSC,H,),S, Molecules Doped in the Diamagnetic Ti(qSC,H,),S, Hosta

/

97.22

/

aThe estimated standard deviations, ug and UT,were calculated from u g = [xi=l(gcalcd - geXptl)*/(n- I)] % and UT = [ x j = l for each least-squares curve withn (Tcalcd ; TeXptl)’/(n observations.

Once the principal values and the orientation of the electron Zeeman and the hyperfine coupling tensors were determined, the theoretical magnetic field was computed for the hyperfine lines of each spectrum. This calculation was performed in order to check the correctness of the data-analysis procedure. Rather than performing an exact calculation which involves the diagonalization of a 16 X 16 matrix, a second-order perturbation treatment incorporating a partitioning technique19 was utilized which requires the diagonalization of two 8 X 8 matrices. A comparison of the experimental line positions with those calculated by this sec-

October 29, 1975

6419 Table 11. Experimental g and T Tensors with Best-Fit Parameters (av) from the SingleCrystal EPR Study of V(+C,H,),S, Doped in Ti(qSC,H,),S, @ =59.05 gx = 1.9964 e =S~.SO gy = 1.9997 $ = -59.52 gz = 1.9689 Tx = (-)66.6 G = 32.41 = (-)111.3 G e = 140.68 z = (-)23.5 G $ = 27.97 For a = -0.963 and h = 0.270, where = a \d,z) + h ldX2Tx = (-)66.6 G P = 87.1 x l o d cm-? T = (-)111.3 G (r-? = 1.86 au = (-)23.5 G x = -1.97 K = 61.5 X 10- c m - I h =30crn-l per cent character of 3dZ2 to 3d,a- 2, a a / b 2= (-0.963)2/(0.270)2 = 12.71;

2

l a

@J

2)